Robot test welding defect automatic detection method

Abstract

The invention discloses a robot test welding defect automatic detection method, which adopts an image analysis algorithm to carry out feature extraction on a preprocessed image, obtains a flatness difference ratio of a welding image, and evaluates the severity of a defect in combination with welding parameter data, so that the intelligent degree of detection is increased, and the accuracy and adaptability of detection are improved. The automatic level of welding defect detection is achieved, meanwhile, through real-time data collection and dynamic adjustment of a detection strategy, different welding environments and parameter changes can be adapted, and the problems that in an existing welding defect detection method, manual visual detection is low in efficiency and is affected by subjective factors are solved; and an automatic detection system with fixed parameters cannot adapt to changes of different welding environments and welding parameters, and the detection accuracy is limited.

Description

Technical Field

[0001] The invention relates to the technical field of welding quality detection, and in particular to an automatic detection method for robot trial welding defects. Background Art

[0002] With the improvement of industrial automation, robot welding technology has been widely used in the manufacturing industry. However, defects that may occur during welding (such as pores, cracks, lack of fusion, etc.) will affect the quality of the welded joint, and then affect the safety and service life of the entire product. Therefore, the detection of welding defects is very important.

[0003] In the prior art, the detection of welding defects mainly relies on manual visual inspection or automatic inspection systems with fixed parameters. Manual visual inspection is inefficient and affected by subjective factors; while the automatic inspection system with fixed parameters cannot adapt to different welding environments and changes in welding parameters, and the detection accuracy is limited. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing welding defect detection methods, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is: to solve the problem that the manual visual detection in the existing welding defect detection method is inefficient and is affected by subjective factors; and the automatic detection system with fixed parameters cannot adapt to different welding environments and changes in welding parameters, and the detection accuracy is limited.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a method for automatic detection of robot trial welding defects, comprising the following steps: S1: data acquisition stage: a. Real-time acquisition of image data of the welding process by a high-resolution camera carried by a welding robot; b. Simultaneous acquisition of welding parameter data during the welding process; S2: defect detection stage: a. Preprocessing the acquired image data to improve image quality; b. Using an image analysis algorithm to perform feature extraction on the preprocessed image to obtain the flatness difference ratio of the welding image; c. Constructing a welding defect degree model in combination with the acquired welding parameter data to obtain the defect degree value under the current image; S3: Evaluating the severity of the defect based on the defect degree value; S4: detection result feedback stage: a. Storing the detected defect degree in a database; b. Feedback the defect degree information to the operator through the user interface to determine whether to perform subsequent processing on the current position correction.

[0008] As a preferred solution of the robot trial welding defect automatic detection method described in the present invention, the welding parameter data collected during the welding process specifically include: current and welding speed.

[0009] As a preferred solution of the robot trial welding defect automatic detection method described in the present invention, the preprocessing of the collected image data specifically includes: image denoising, image enhancement, image binarization and image standardization.

[0010] As a preferred solution of the robot trial welding defect automatic detection method described in the present invention, wherein: the image analysis algorithm is used to extract features from the preprocessed image, and obtaining the flat difference ratio of the welding image specifically includes the following steps: H1: obtaining a two-dimensional point cloud data map of the current image after obtaining the welding image; H2: selecting point cloud data with a darker grayscale in the two-dimensional point cloud data map to form a point cloud area representing welding defects; H3: obtaining the area occupied by the point cloud area; H4: synchronously obtaining the overall area size of the two-dimensional point cloud data map; H5: obtaining the flat difference ratio of the welding image according to the following formula;

[0011]

[0012] Where δ represents the flatness difference ratio of the welding image, %; S 深 Indicates the area occupied by the point cloud region obtained in H3, m 2 ; S 总 Indicates the overall area obtained in H4, m 2 .

[0013] As a preferred solution of the robot trial welding defect automatic detection method of the present invention, the constructed welding defect degree model is specifically:

[0014]

[0015] Among them, η is the defect degree value; δ represents the flatness difference ratio of the welding image, %; I is the current of the current welding process, mA; v is the welding speed of the current welding process, mm / s; 1.36 and -1.07 are both adjustment constants; dx is the integration constant.

[0016] As a preferred solution of the robot trial welding defect automatic detection method described in the present invention, wherein: when the defect severity is evaluated based on the defect degree value: when the defect degree value is higher than the threshold value, it is defined that the defects in the current image welding do not meet the standards, and degree feedback and correction are required.

[0017] As a preferred solution of the robot trial welding defect automatic detection method described in the present invention, the threshold is specifically set to 3.78 or 3.791.

[0018] The beneficial effects of the present invention are as follows: the present invention provides a method for automatic detection of trial welding defects by a robot, which uses an image analysis algorithm to extract features from preprocessed images, obtains the flatness difference ratio of the welding image, and evaluates the severity of the defects in combination with welding parameter data, thereby increasing the intelligence of the detection, improving the accuracy and adaptability of the detection, and realizing the automation level of welding defect detection. At the same time, the present invention can adapt to different welding environments and parameter changes by real-time data collection and dynamic adjustment of the detection strategy, thereby solving the problems in the existing welding defect detection methods that artificial visual detection is inefficient and is affected by subjective factors; and the automatic detection system with fixed parameters cannot adapt to different welding environments and changes in welding parameters, and has limited detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0020] Figure 1 A method flow chart of the automatic detection method for robot trial welding defects provided by the present invention.

[0021] Figure 2 A flow chart of a method for extracting features from a preprocessed image using an image analysis algorithm to obtain a flatness difference ratio of a welding image provided by the present invention. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0023] In the prior art, the detection of welding defects mainly relies on manual visual inspection or automatic inspection systems with fixed parameters. Manual visual inspection is inefficient and affected by subjective factors; while the automatic inspection system with fixed parameters cannot adapt to different welding environments and changes in welding parameters, and the detection accuracy is limited.

[0024] Therefore, please refer to Figure 1 The present invention provides a method for automatically detecting defects in a robot trial welding, comprising the following steps:

[0025] S1: Data collection phase:

[0026] a. The high-resolution camera on the welding robot collects image data of the welding process in real time;

[0027] b. Collect welding parameter data during welding process at the same time;

[0028] S2: Defect detection stage:

[0029] a. Preprocess the collected image data to improve image quality;

[0030] b. Use image analysis algorithm to extract features from the preprocessed image and obtain the flatness difference ratio of the welding image;

[0031] c. Combine the collected welding parameter data to build a welding defect degree model and obtain the defect degree value under the current image;

[0032] S3: Defect severity assessment based on defect severity value;

[0033] S4: Test result feedback stage:

[0034] a. Store the detected defect levels in the database;

[0035] b. Feedback the defect degree information to the operator through the user interface to determine whether to correct the current position.

[0036] It should be noted that the welding parameter data collected during the welding process specifically include: current and welding speed. When the robot is performing basic welding, the welding parameter data are all fixed values ​​and can be directly read.

[0037] Specifically, the preprocessing of the collected image data includes: image denoising, image enhancement, image binarization and image standardization.

[0038] It should be noted that:

[0039] 1. Image denoising:

[0040] Cause: Smoke, sparks, etc. generated during welding may cause noise in the image.

[0041] Method: Use median filtering, mean filtering or Gaussian filtering to remove random noise in the image.

[0042] 2. Image enhancement:

[0043] Cause: Welding images may be affected by problems such as uneven lighting and insufficient contrast, which may affect the identification of defects.

[0044] Methods: Histogram equalization is used to improve the contrast of the image; gamma correction is applied to adjust the brightness of the image; and a local contrast enhancement algorithm is used to enhance the details of the welding area.

[0045] 3. Image Binarization:

[0046] Reason: Converting an image into a binary image can simplify subsequent processing.

[0047] Method: Use global or local thresholding methods, such as Otsu's algorithm, to convert the image to black and white.

[0048] 4. Image standardization:

[0049] Reason: Ensure that image data input to deep learning models has a uniform format and range.

[0050] Method: Normalize the image data to the range of [0,1] or [-1,1]; resize the image to meet the requirements of the model input.

[0051] For further information, see Figure 2 , the image analysis algorithm is used to extract features from the preprocessed image, and obtaining the flatness difference ratio of the welding image specifically includes the following steps:

[0052] H1: Get the 2D point cloud data of the current image after getting the welding image;

[0053] It should be noted that the operation of obtaining a two-dimensional point cloud data map is a mature application of existing technology;

[0054] The following are the general steps to obtain a 2D point cloud data map from a welding image:

[0055] Image preprocessing:

[0056] The welding images are preprocessed by denoising, contrast enhancement, histogram equalization and other operations to improve the image quality.

[0057] 2. Image segmentation:

[0058] Image segmentation technology is used to separate the welding area from the background. This can be achieved through threshold segmentation, edge detection, region growing, cluster analysis and other methods.

[0059] Feature extraction:

[0060] Extract feature points in the segmented welding area. These feature points can be corner points, edge points or other significant points. Commonly used feature extraction methods include SIFT (Scale Invariant Feature Transform), SURF (Speeded Robust Features), ORB (Oriented FAST and Rotated BRIEF), etc.

[0061] 2D point cloud generation:

[0062] The extracted feature point set is converted into a two-dimensional point cloud data map. Each point usually contains its coordinates (x, y) in the image and possible other attributes such as intensity, color or texture information.

[0063] Here are the specific steps:

[0064] a. Denoising:

[0065] Apply filters (e.g., median filter, Gaussian filter) to remove noise from the image.

[0066] b. Contrast enhancement and histogram equalization:

[0067] Adjust the contrast of the image to make the details of the weld area clearer.

[0068] Use histogram equalization to improve the dynamic range of the image and enhance the visibility of the weld area.

[0069] c. Split welding area:

[0070] Apply adaptive threshold segmentation or Otsu's method to separate the weld area from the background.

[0071] For complex backgrounds, it may be necessary to use more advanced segmentation techniques such as region growing or graph-based segmentation.

[0072] d. Edge detection and corner detection:

[0073] Use the Canny edge detector or other edge detection algorithms to find the edges of the weld area.

[0074] The corners in the weld area are located by a corner detection algorithm (such as Harris corner detection).

[0075] e. Point cloud generation:

[0076] Generate 2D point cloud data based on the detected feature points. Each point can be represented by a structure or class, which contains the coordinates and other properties of the point.

[0077] f. Point cloud post-processing:

[0078] The generated two-dimensional point cloud is post-processed, such as removing outliers, filtering, clustering, etc., to improve the quality of the point cloud.

[0079] Through the above steps, a 2D point cloud data map can be obtained from the welding image, which is very useful for subsequent defect detection, analysis and robot path planning tasks. The 2D point cloud data map provides a way to represent the characteristics of the welding area on the image plane, which is helpful for further processing and analysis.

[0080] H2: Select the point cloud data with darker grayscale in the two-dimensional point cloud data map to form a point cloud area representing welding defects;

[0081] H3: Get the area occupied by the point cloud area and directly perform area statistics;

[0082] H4: Synchronously obtain the overall area size of the 2D point cloud data map and directly perform area statistics;

[0083] H5: Obtain the flatness difference ratio of the welding image according to the following formula;

[0084]

[0085] Where δ represents the flatness difference ratio of the welding image, %; S 深 Indicates the area occupied by the point cloud region obtained in H3, m 2 ; S 总 Indicates the overall area obtained in H4, m 2 .

[0086] Furthermore, the constructed welding defect degree model is specifically as follows:

[0087]

[0088] Among them, η is the defect degree value; δ represents the flatness difference ratio of the welding image, %; I is the current of the current welding process, mA; v is the welding speed of the current welding process, mm / s; 1.36 and -1.07 are both adjustment constants; dx is the integration constant.

[0089] Furthermore, when evaluating the severity of defects based on the defect degree value: when the defect degree value is higher than the threshold value, it is defined that the defects in the current image welding do not meet the standards, and degree feedback and correction are required.

[0090] Specifically, the threshold is set to 3.78 or 3.791.

[0091] In order to verify the excellent technical effects of the present invention, the following experimental verification process is designed:

[0092] Test verification process:

[0093] 1. Preparation stage:

[0094] Select representative welding samples covering different welding environments and parameter variations.

[0095] Prepare the equipment required for the test, including welding robots, high-resolution cameras, data acquisition systems, etc.

[0096] Determine the welding parameters for the test, such as current, welding speed, etc.

[0097] 2. Data collection stage:

[0098] Use the high-resolution camera on the welding robot to collect image data of the welding process in real time.

[0099] At the same time, welding parameter data such as current and welding speed are collected during the welding process.

[0100] 3. Image preprocessing:

[0101] The collected image data is preprocessed, including denoising, enhancement, binarization, standardization, etc., to improve image quality.

[0102] 4. Defect detection and analysis:

[0103] Image analysis algorithm is applied to extract features of the preprocessed image and obtain the flatness difference ratio of the welding image.

[0104] Combined with the collected welding parameter data, the constructed welding defect degree model is used to evaluate the defect severity.

[0105] 5. Result evaluation and feedback:

[0106] The severity of the defect is evaluated based on the defect severity value and compared with the preset threshold.

[0107] The degree of detected defects is stored in the database and fed back to the operator through the user interface.

[0108] 6. Repeat the experiment and data analysis:

[0109] Repeat the above process several times to ensure the reliability and repeatability of the data.

[0110] The collected test data were statistically analyzed to evaluate the performance of the present invention.

[0111] Test verification data table:

[0112] The following is a sample test validation data table:

[0113]

[0114]

[0115] The present invention provides a method for automatically detecting defects in trial welding by a robot. The method uses an image analysis algorithm to extract features from a preprocessed image, obtains a flat difference ratio of the welding image, and evaluates the severity of the defect in combination with welding parameter data, thereby increasing the intelligence of the detection, improving the accuracy and adaptability of the detection, and realizing the automation level of welding defect detection. At the same time, the present invention can adapt to different welding environments and parameter changes by real-time data collection and dynamic adjustment of the detection strategy, thereby solving the problems in the existing welding defect detection methods that artificial visual detection is inefficient and is affected by subjective factors; and the automatic detection system with fixed parameters cannot adapt to different welding environments and changes in welding parameters, and has limited detection accuracy.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A robot test welding defect automatic detection method, characterized in that: The steps include: S1: Data collection phase: a. The high-resolution camera on the welding robot collects image data of the welding process in real time; b. Collect welding parameter data during welding process at the same time; S2: Defect detection stage: a. Preprocess the collected image data to improve image quality; b. Use image analysis algorithm to extract features from the preprocessed image and obtain the flatness difference ratio of the welding image; c. Combine the collected welding parameter data to build a welding defect degree model and obtain the defect degree value under the current image; S3: Evaluate the severity of the defect based on the defect severity value; S4: Test result feedback stage: a. Store the detected defect levels in the database; b. Feedback the defect degree information to the operator through the user interface to determine whether to correct the current position.

2. The robot trial welding defect automatic detection method according to claim 1 is characterized in that: The welding parameter data collected during the welding process specifically include: current and welding speed.

3. The robot trial welding defect automatic detection method according to claim 2 is characterized in that: The collected image data is preprocessed including: image denoising, image enhancement, image binarization and image standardization.

4. The method for automatic detection of robot trial welding defects according to claim 3 is characterized in that: The image analysis algorithm is used to extract features from the preprocessed image, and obtaining the flatness difference ratio of the welding image specifically includes the following steps: H1: Get the 2D point cloud data of the current image after getting the welding image; H2: Select the point cloud data with darker grayscale in the two-dimensional point cloud data image to form a point cloud area representing welding defects; H3: Get the area occupied by the point cloud area; H4: synchronously obtain the overall area size of the two-dimensional point cloud data map; H5: Obtain the flatness difference ratio of the welding image according to the following formula; Where δ represents the flatness difference ratio of the welding image, %; S 深 Indicates the area occupied by the point cloud region obtained in H3, m 2 ; S 总 Indicates the overall area obtained in H4, m 2 .

5. The method for automatic detection of robot trial welding defects according to claim 4 is characterized in that: The welding defect degree model constructed is specifically: Among them, η is the defect degree value; δ represents the flatness difference ratio of the welding image, %; I is the current of the current welding process, mA; v is the welding speed of the current welding process, mm / s; 1.36 and -1.07 are both adjustment constants; dx is the integration constant.

6. The method for automatic detection of robot trial welding defects according to claim 5 is characterized in that: When the defect severity is evaluated based on the defect severity value: when the defect severity value is higher than a threshold value, it is defined that the defect during welding of the current image does not meet the standard, and degree feedback and correction are required.

7. The method for automatic detection of robot trial welding defects according to claim 6 is characterized in that: The threshold is specifically set to 3.78 or 3.791.

Images